Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/adilkalam/orca/refinegit clone --depth 1 https://github.com/adilkalam/orcaWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00087 | $0.00987 |
| Opus 5 | $0.00044 | $0.00494 |
| Sonnet 5 | $0.00017 | $0.00197 |
| Haiku 4.5 | $0.00009 | $0.00099 |
Grade A, and why
refine scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
This is a copy
86% identical to fortify — 49 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 69 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/refine — Design Refinement Router (individual commands)
Tweak entry for the refinement verbs. Each flag is an individual command run OUTSIDE the /impeccable
pipeline — so there is no separate validator agent, and the cognition constraint loop IS the
enforcement (mandatory). Zero loop logic of its own (#POISON_PATH duplication): the loop is defined
once at ~/.claude/docs/reference/cognition-constraint-loop.md.
Parse flag (exactly one required)
| Flag | Verb / craft skill |
|---|---|
--animate |
animate (light motion; heavy motion → /motion-design) |
--bolder |
bolder |
--colorize |
colorize |
--delight |
delight |
--layout |
layout |
--overdrive |
overdrive |
--quieter |
quieter |
--typeset |
typeset |
If no flag or an unrecognized flag: print this table, ask the user to pick. Do not guess. Do not run.
Run — the in-thread cognition constraint loop
Target routing (#PATH_DECISION — one rule, not 5 copies): apply
~/.claude/docs/concepts/ios-design-contract/target-routing.md. If the TARGET ends in .swift, load
Skill("ios-impeccable-hub") (instead of impeccable-hub) and use the Swift detector below; otherwise
keep the CSS path unchanged.
Run the loop at ~/.claude/docs/reference/cognition-constraint-loop.md with VERB = <flag>, loading
Skill("impeccable-hub") (the aesthetic; Skill("ios-impeccable-hub") for a .swift target — see routing
above) + Skill("<flag>") (craft spine):
- R1 BIND — cognition
checkpointemits the typed FORBIDDEN/FORWARD constraints this verb-on-this- target can trip (detector/banned-rule ids); capture the returned ids. - R2 WORK — edit the target in-thread under the bound ids, applying the craft + the user's critique
verbatim (FR-6). Before writing styling code, read the relevant doctrine: the CSS manifesto
(
~/.claude/docs/concepts/llm-css-manifesto.md) for web; the loadedios-impeccable-hubfor a.swifttarget (per the routing rule above). - EVALUATE — check every bound constraint + run the detector self-check. For a
.swifttarget:/Users/adilkalam/ORCA-OS/mcp/swift-design-detector/bin/swiftdesigncheck detect --json <file>; otherwisenode /Users/adilkalam/ORCA-OS/mcp/design-detector/bin/designcheck.js detect --json <file>. Both:EXIT=2= findings present. Record a cognitionthought(per id: satisfied|unsatisfied). - R(n) LOOP — fix + re-evaluate until none unsatisfied. MAX N=2, then escalate to the user. You may NOT claim done with an open constraint.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 69 lines · 87 tokens per session scan A 757a4c1d2948
refine is a command published in the GitHub repository adilkalam/orca (2 stars, last pushed 1mo ago), licensed MIT. It adds 87 tokens to every session and 987 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. It is 86% identical to fortify, differing in 49 lines, and is treated as a copy.
Other commands, from other repositories
git
Git operations with intelligent commit messages and workflow optimization.
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
analyze
Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.
converge
Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.